A method and system for constructing a digital twin of a terrestrial ecosystem

Abstract

The present application relates to a kind of digital twin construction method and system of terrestrial ecosystem, the method includes: according to the biological and environmental data collected in ecological protection area, obtain plant and animal information and environmental quality assessment situation;Based on the evaluation model, the plant and animal information, environmental quality assessment situation, biological feature vector generated based on biological data, environmental feature vector generated based on environmental data and the comprehensive feature vector generated by the fusion of two kinds of vectors are processed, and the ecosystem service function evaluation result is obtained;Based on plant and animal information, environmental quality assessment situation, ecosystem service function evaluation result generates comprehensive feature matrix, carries out deep learning to the comprehensive feature matrix and constructs digital twin engine;After digital twin engine and model static base information are based on and digital twin big model is constructed, in response to the real-time data of input digital twin big model, the linkage state of species, environment and ecosystem service function is rendered in visual interface.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ecology, and in particular to a method and system for constructing a digital twin of a terrestrial ecosystem. BACKGROUND

[0002] With the improvement of ecological protection awareness, monitoring of the ecological system has become the key to maintaining biodiversity and ecological balance. Real-time monitoring of the ecological system can provide early warning of environmental degradation, assessment of ecological service functions, and provide a basis for ecological restoration and sustainable management.

[0003] With the rapid development of digital technology, digital twin technology has gradually emerged and been applied in many fields. A digital twin is a virtual digital model corresponding to a physical entity, which is constructed by integrating physical models, sensor data and historical data and other multi-source information, and can reflect the state of the physical entity in real time. In the field of ecology, although there have been attempts to construct digital twins of ecological systems to monitor the ecological system, there are still many challenges.

[0004] 1. The data collection is single, mostly focusing on the collection of environmental data, ignoring the importance of biological data, resulting in incomplete assessment of the ecological system function; 2. Due to the wide range of ecological systems and complex environment, there are problems such as difficulty in data collection and poor real-time performance; 3. At the data fusion level, different modal data often have different data structures and time resolutions, and simple data superposition cannot fully utilize the advantages of multi-source data, making it difficult to obtain comprehensive data that accurately reflect the state of the ecological system; 4. Data collaboration in the fields of ecology, geographic information, computer science and other fields is difficult, and it is difficult to form a sustainable digital twin ecological circle.

[0005] Therefore, there are still many problems to be solved in the application of digital twin technology in the ecological field in the prior art, and a feasible scheme for constructing a digital twin of an ecological system is urgently needed. SUMMARY

[0006] The present application aims to provide a method and system for constructing a digital twin of a terrestrial ecosystem to solve the problems existing in the prior art. The technical problems to be solved by the present application are solved by the following technical solutions.

[0007] In a first aspect, the present application provides a method for constructing a digital twin of a terrestrial ecosystem, comprising:

[0008] According to the biological data and environmental data collected in the ecological protection area, the information of animals and plants and the environmental quality assessment situation in the ecological protection area are obtained, the biological data are used to generate a biological feature vector for animal and plant identification, and the environmental data are used to generate an environmental feature vector for environmental quality assessment;

[0009] Based on the evaluation of the large model to the animal and plant information, the environment quality evaluation situation, the biological feature vector, the environment feature vector and the comprehensive feature vector are processed, and the ecosystem service function evaluation result of the ecological protection area is obtained, and the comprehensive feature vector is determined based on the fusion of the biological feature vector, the environment feature vector and the dynamically adjusted biological weight and environment weight;

[0010] Based on the animal and plant information, the environment quality evaluation situation and the ecosystem service function evaluation result, a comprehensive feature matrix is generated, and a digital twin engine is constructed by deep learning of the comprehensive feature matrix; wherein the comprehensive feature matrix is composed of fusion feature vectors of multiple space-time units arranged in space-time sequence, and the fusion feature vector of a single space-time unit is formed by multi-modal feature splicing of the animal and plant information, the environment quality evaluation situation and the ecosystem service function evaluation result corresponding to the current space-time unit;

[0011] After constructing the digital twin large model based on the digital twin engine and the model static base information, in response to inputting real-time data of the digital twin large model, the linkage state of species, environment and ecosystem service function is rendered in a visual interface, and the visual interface is a virtual reality (VR) interface or an augmented reality (AR) interface.

[0012] In a second aspect, the embodiments of the present application provide a digital twin construction system of a terrestrial ecosystem, comprising:

[0013] An acquisition module is configured to acquire animal and plant information and environment quality evaluation situation in the ecological protection area according to biological data and environment data collected in the ecological protection area, wherein the biological data is used to generate a biological feature vector for animal and plant identification, and the environment data is used to generate an environment feature vector for environment quality evaluation;

[0014] A processing acquisition module is configured to process the animal and plant information, the environment quality evaluation situation, the biological feature vector, the environment feature vector and the comprehensive feature vector based on an evaluation large model, and obtain an ecosystem service function evaluation result of the ecological protection area, wherein the comprehensive feature vector is determined based on the fusion of the biological feature vector, the environment feature vector and the dynamically adjusted biological weight and environment weight;

[0015] The generating module is configured to generate a comprehensive feature matrix based on the animal and plant information, the environment quality assessment condition, and the ecosystem service function assessment result, and to construct a digital twin engine through deep learning of the comprehensive feature matrix; wherein the comprehensive feature matrix is formed by fusion feature vectors of multiple space-time units arranged in a space-time sequence, and the fusion feature vector of a single space-time unit is formed by multi-modal feature splicing of the animal and plant information, the environment quality assessment condition, and the ecosystem service function assessment result corresponding to the current space-time unit.

[0016] The rendering module is configured to, after constructing a digital twin large model based on the digital twin engine and model static base information, render the linkage state of the species, the environment, and the ecosystem service function in a visual interface in response to input of real-time data of the digital twin large model, wherein the visual interface is a virtual reality (VR) interface or an augmented reality (AR) interface.

[0017] The technical scheme provided by the embodiments of the present application can comprehensively analyze the biological feature vector generated based on biological data, the environmental feature vector generated based on environmental data, the comprehensive feature vector generated by fusing the biological weight and the environmental weight dynamically adjusted based on the two types of vectors, the animal and plant information in the ecological protection zone, and the environment quality assessment condition, to obtain a more comprehensive, accurate, and scientific ecosystem service function assessment result. The comprehensive feature matrix generated based on the animal and plant information, the environment quality assessment condition, and the ecosystem service function assessment result can improve the coupling modeling precision between ecological variables. After obtaining the comprehensive feature matrix, the digital twin engine is constructed through deep learning of the comprehensive feature matrix, the digital twin large model is constructed based on the digital twin engine and the model static base information, the real-time data input to the digital twin large model is processed, and the linkage state of the species, the environment, and the ecosystem service function is rendered in the VR interface or the AR interface, so that the state of the physical ecosystem and the digital twin large model can be synchronized, the timeliness and accuracy of the large model can be ensured, and the content provided by the digital twin large model can be visually displayed through VR technology or AR technology, thereby providing an immersive user experience. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 A digital twin construction method for a terrestrial ecosystem provided by the embodiments of the present application is shown in the following schematic diagram.

[0019] Figure 2 A simplified flowchart showing the processing of real-time data by the digital twin large model provided by the embodiments of the present application is shown in the following diagram.

[0020] Figure 3The overall implementation flowchart of the digital twin construction of the embodiment of the present application is shown.

[0021] Figure 4 The digital twin construction system schematic diagram of the terrestrial ecosystem provided by the embodiment of the present application is shown. DETAILED DESCRIPTION

[0022] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0023] The embodiment of the present application provides a digital twin construction method of a terrestrial ecosystem, as shown in the following steps. Figure 1

[0024] Step 101, according to the biological data and environmental data collected in the ecological protection zone, the information of animals and plants and the environmental quality evaluation situation in the ecological protection zone are obtained, the biological data is used to generate a biological feature vector for animal and plant identification, and the environmental data is used to generate an environmental feature vector for environmental quality evaluation.

[0025] After selecting a certain ecological protection zone, the embodiment of the present application uses a collection device (such as an unmanned device equipped with various sensors) to collect biological data and environmental data in the ecological protection zone within one or more specific time periods. After obtaining the biological data and environmental data through data collection, a biological feature vector is generated based on the biological data, and an environmental feature vector is generated based on the environmental data. The generated biological feature vector is used to indicate the global biological features in the ecological protection zone, and the generated environmental feature vector is used to indicate the global environmental features in the ecological protection zone.

[0026] Based on the biological feature vector, the animals and plants in the ecological protection zone are identified, and based on the environmental feature vector, the environmental quality in the ecological protection zone is evaluated, so as to obtain the information of animals and plants and the environmental quality evaluation situation in the ecological protection zone. The obtained information of animals and plants is the global information of animals and plants in the ecological protection zone, and the obtained environmental quality evaluation situation is the global environmental quality evaluation situation in the ecological protection zone. Through the identification of the species in the ecological protection zone, the information such as the species, quantity and distribution of the animals and plants in the ecological protection zone can be understood. Through the environmental quality evaluation of the ecological protection zone, it is beneficial to timely find the environmental problems in the ecological protection zone, and to provide a scientific basis for ecological protection and management.

[0027] ​In step 102, the evaluation large model is used to process the biological and environmental information, the environmental quality evaluation, the biological feature vector, the environmental feature vector, and the comprehensive feature vector to obtain the ecosystem service function evaluation result of the ecological protection zone. The comprehensive feature vector is determined based on the biological feature vector, the environmental feature vector, and the dynamically adjusted biological weight and environmental weight.

[0028] After obtaining the biological and environmental information in the ecological protection zone based on the biological feature vector and the environmental quality evaluation in the ecological protection zone based on the environmental feature vector, the evaluation large model is used to process the biological feature vector, the environmental feature vector, the comprehensive feature vector, the biological and environmental information in the ecological protection zone, and the environmental quality evaluation to evaluate the ecosystem service function of the ecological protection zone and obtain the ecosystem service function evaluation result.

[0029] The comprehensive feature vector is determined based on the biological feature vector, the environmental feature vector, and the dynamically adjusted biological weight and environmental weight. The biological weight and the environmental weight can be dynamically adjusted based on the real-time contribution of biological factors and environmental factors, such as adjusting the corresponding weight based on the actual contribution of biological factors and environmental factors to the ecosystem health and function at the current time point. The biological feature vector reflects information such as biodiversity and species richness, and the environmental feature vector covers multiple environmental factors. By integrating the biological feature vector and the environmental feature vector and combining the dynamically adjusted biological weight and environmental weight to generate the comprehensive feature vector, the ecological condition of the ecological protection zone can be more comprehensively evaluated.

[0030] For the biological feature vector and the environmental feature vector, they respectively contain biological information and environmental information in the ecological protection zone. The two vectors provide basic data, and through analysis of these data, more specific biological information (mainly referring to plants and animals) and environmental quality evaluation can be extracted. The plant and animal information obtained based on the analysis of the biological feature vector belongs to higher-level information, such as species diversity, distribution of key species, etc. The environmental quality evaluation obtained based on the analysis of the environmental feature vector also belongs to higher-level information, such as water quality compliance, soil pollution degree, etc. Although the plant and animal information and the environmental quality evaluation provide important information, they are the results obtained after preliminary analysis and may lose some details and potential relationships in the original data. When considering the plant and animal information and the environmental quality evaluation in the ecological protection zone, analyzing the biological feature vector, the environmental feature vector, and the comprehensive feature vector can dig out more potential information and mutual relationships.

[0031] By comprehensively analyzing the biological feature vector, the environmental feature vector, the comprehensive feature vector, the information of animals and plants in the ecological protection zone, the environmental quality assessment situation, the complex interaction between the biological and environmental can be revealed, and the overall condition of the ecological system can be more comprehensively, more accurately and more deeply understood, and then the ecosystem service function can be more comprehensively, more accurately and more deeply evaluated, and the ecosystem service function evaluation result can be obtained.

[0032] In step 103, a comprehensive feature matrix is generated based on the information of animals and plants, the environmental quality assessment situation and the ecosystem service function evaluation result, and a digital twin engine is constructed by deep learning on the comprehensive feature matrix; wherein the comprehensive feature matrix is composed of fusion feature vectors of multiple spatio-temporal units arranged in spatio-temporal sequence, and the fusion feature vector of a single spatio-temporal unit is formed by multi-modal feature splicing of the information of animals and plants, the environmental quality assessment situation and the ecosystem service function evaluation result corresponding to the current spatio-temporal unit.

[0033] After obtaining the information of animals and plants in the ecological protection zone, the environmental quality assessment situation, and using the evaluation large model to obtain the ecosystem service function evaluation result, the related data in each spatio-temporal unit of the ecological protection zone is processed. The spatio-temporal unit refers to a specific region and time period divided according to the two dimensions of space and time in the ecological protection zone. For each spatio-temporal unit, the comprehensive feature vector is processed with the information of animals and plants, the environmental quality assessment situation and the ecosystem service function evaluation result corresponding to the spatio-temporal unit to determine the standard data, multi-modal features (at least including biological features, environmental features and service function features) are extracted in the standard data, and each modal feature vector is generated based on the extracted features; in the same time and space dimension, the modal feature vectors are spliced to generate the fusion feature vector of the current spatio-temporal unit. For example, for each spatio-temporal unit (such as 10m grid x 1h time slice), the biological features, environmental features and service function features extracted in the spatio-temporal unit are generated into multi-modal feature vectors and vector spliced to obtain the fusion feature vector, and the fusion feature vectors of multiple spatio-temporal units are arranged in time sequence and space grid order to form a comprehensive feature matrix.

[0034] After obtaining the comprehensive feature matrix, a digital twin engine is constructed by deep learning on the comprehensive feature matrix. The digital twin engine includes a dynamic calculation layer and a rule engine layer, which are respectively responsible for real-time calculation and logical decision, and together constitute the core function of the digital twin engine; in the hierarchical structure, the dynamic calculation layer is usually located at the bottom, responsible for basic calculation, and the rule engine layer is located at the upper layer, responsible for logical judgment and decision support.

[0035] In constructing the digital twin engine based on the comprehensive feature matrix, the comprehensive feature matrix is taken as the input, a dynamic calculation layer is formed through deep learning, the dynamic calculation layer is taken as the core, a rule engine layer is integrated, and a digital twin engine is encapsulated. The constructed digital twin engine has the capabilities of real-time data access, state calculation, rule judgment and response triggering, is the core driving module of the digital twin large model, and supports dynamic operation and intelligent decision of the large model.

[0036] In step 104, after constructing the digital twin large model based on the digital twin engine and the model static base information, the linkage state of the species, the environment and the ecosystem service function is rendered on the visual interface in response to the real-time data input into the digital twin large model, and the visual interface is a virtual reality (VR) interface or an augmented reality (AR) interface.

[0037] The digital twin engine constructed based on deep learning of the comprehensive feature matrix is the core calculation and decision module of the digital twin large model, and is responsible for real-time data processing, state calculation, rule response and other dynamic functions; the model static base information provides the spatial structure and basic attributes of the large model, including, for example, terrain, vegetation type, soil texture and other information, which provides a spatial reference framework for the digital twin engine, so that the calculation results can be accurately positioned and displayed in geographic space.

[0038] The digital twin engine is responsible for calculation and response, the model static base information is the skeleton, which can provide spatial structure and basic attributes, and the digital twin engine and the model static base information are combined to construct the digital twin large model.

[0039] After constructing the digital twin large model, real-time data is input into the digital twin large model, and the input real-time data is processed by the digital twin large model. For example, through the Internet of Things interface, the real-time acquired plant and animal information, environmental quality evaluation and ecosystem service function evaluation results are connected to the digital twin engine, and the model state of the digital twin large model is dynamically updated by the digital twin engine, so that the large model is synchronized with the real ecosystem.

[0040] After the digital twin large model processes the input real-time data, the linkage state of the species, the environment and the ecosystem service function is output, and the output content is visualized through VR (Virtual Reality) or AR (Augmented Reality) technology, so that users can intuitively understand the state of the ecosystem through the VR interface or the AR interface.

[0041] The latest data is continuously acquired through the Internet of Things and processed by the digital twin large model, ensuring that the large model always reflects the current state of the ecosystem and maintains its timeliness and accuracy. The output of the digital twin large model is displayed through a visual interface. Abstract data can be converted into intuitive three-dimensional images or augmented reality images through VR or AR technology to ensure a good visual experience for users.

[0042] The above embodiments of the present application analyze biological data and environmental data to obtain information about animals and plants in the ecological protection zone and environmental quality assessment, generate a biological feature vector based on biological data, an environmental feature vector based on environmental data, a comprehensive feature vector generated by fusing the biological feature vector and the environmental feature vector, and a biological weight and an environmental weight dynamically adjusted based on the two types of vectors, and comprehensively analyze the information about animals and plants in the ecological protection zone and the environmental quality assessment to obtain a more comprehensive, accurate, and scientific ecological system service function evaluation result. A comprehensive feature matrix is generated based on the information about animals and plants, the environmental quality assessment, and the ecological system service function evaluation result, which can improve the coupling modeling accuracy between ecological variables. After obtaining the comprehensive feature matrix, a digital twin engine is constructed through deep learning of the comprehensive feature matrix, a digital twin large model is constructed based on the digital twin engine and model static base information, real-time data input into the digital twin large model is processed, and the linkage state of species, environment, and ecological system service function is rendered in a VR interface or an AR interface. This can achieve state synchronization between the physical ecosystem and the digital twin large model, ensure the timeliness and accuracy of the large model, and provide an immersive user experience through visual display of the content provided by the digital twin large model through VR technology or AR technology.

[0043] The following describes a scheme for obtaining information about animals and plants in an ecological protection zone and environmental quality assessment, and obtaining an ecological system service function evaluation result of the ecological protection zone. When biological data and environmental data collected in an ecological protection zone are used to obtain information about animals and plants in the ecological protection zone and environmental quality assessment, the following steps are included:

[0044] A biological feature vector is generated based on the collected biological data, and an environmental feature vector is generated based on the collected environmental data. The biological data at least includes image data, audio data, and positioning data of animals and plants, and the environmental data at least includes meteorological data and physical environmental data. A biological large model is used to identify species based on the biological feature vector to obtain information about animals and plants in the ecological protection zone, and an environmental large model is used to analyze the environment based on the environmental feature vector to obtain environmental quality assessment in the ecological protection zone.

[0045] The biological data collected by the collection device at least includes image data, audio data and positioning data of animals and plants, and the collected environmental data at least includes meteorological data and physical environment data. For example, the environmental data includes but is not limited to temperature, humidity, soil condition, water quality, light intensity, air pressure, wind speed, seasonal factors, etc. By collecting environmental data, background information of animals and plants living and moving in the ecological protection zone can be provided to more comprehensively describe the regional characteristics of the ecological protection zone in combination with biological data.

[0046] After generating the biological feature vector based on the collected biological data, the biological feature vector is subjected to species recognition based on a pre-constructed biological large model to obtain animal and plant information. The biological large model is a species recognition model, and the biological feature vector is subjected to inference analysis based on the biological large model to obtain animal and plant information output by the biological large model, so as to effectively recognize the species in the ecological protection zone and obtain global animal and plant information. After generating the environmental feature vector based on the collected environmental data, the environmental feature vector is subjected to environmental quality evaluation based on a pre-constructed environmental large model. The environmental large model is an environmental evaluation model, and the environmental feature vector is analyzed based on the environmental large model to obtain global environmental quality evaluation of the ecological protection zone output by the environmental large model. The environmental quality evaluation output by the environmental large model includes, for example, environmental quality score, which is determined based on comprehensive evaluation of environmental factors such as water quality, soil, and meteorological conditions in the ecological protection zone.

[0047] Optionally, when the biological feature vector, the environmental feature vector, the comprehensive feature vector, the animal and plant information, and the environmental quality evaluation are processed based on the evaluation large model to obtain the ecosystem service function evaluation result of the ecological protection zone, the following steps are included:

[0048] The biological feature vector and the environmental feature vector are combined to determine a comprehensive feature vector based on the dynamically adjusted biological weight and environmental weight. The biological feature vector, the environmental feature vector, the comprehensive feature vector, the animal and plant information, and the environmental quality evaluation are subjected to comprehensive inference analysis based on the optimized evaluation large model to generate the ecosystem service function evaluation result.

[0049] The comprehensive feature vector is determined based on the biological feature vector, the environmental feature vector, and the dynamically adjusted biological weight and environmental weight. By integrating the biological feature vector and the environmental feature vector to form the comprehensive feature vector, the health status of the ecological protection area can be more comprehensively evaluated. Moreover, the formation of the comprehensive feature vector is not only based on the biological feature vector and the environmental feature vector, but also takes into account the dynamically adjusted biological weight and environmental weight. The biological weight and the environmental weight are dynamically adjusted based on the contribution of biological data and environmental data to the evaluation of ecosystem service functions. This dynamic adjustment mechanism can flexibly adjust the weights according to the characteristics and protection goals of different ecological regions, so as to obtain the comprehensive feature vector in a targeted manner. Compared with the prior art, which has difficulty in integrating multi-source heterogeneous data (such as biological data and environmental data) and has fixed weights that cannot be dynamically adjusted, in the embodiments of the present application, biological data (such as images, audio, positioning) and environmental data (such as weather, soil, water quality) are integrated to generate biological feature vectors and environmental feature vectors, and a comprehensive feature vector is generated based on dynamically adjusted weights, solving the problem of data integration, and using a dynamic weight adjustment mechanism to flexibly adjust the weights according to the real-time contribution of the data to the evaluation of ecosystem service functions, so as to more accurately reflect the overall status of the ecosystem.

[0050] Exemplarily, a specific example of dynamically adjusting the comprehensive feature vector is given here:

[0051] 1. Input data and initial feature vector

[0052] Biological feature vector (dimension = 4): B = [0.2, 0.95, 0.15, 0.5] (representing the number of Siberian tigers, confidence, the number of elk, and activity range); Environmental feature vector (dimension = 4): E = [0.6, 0.65, 0.68, 0.67] (representing temperature, humidity, soil pH, and snow depth).

[0053] 2. Dynamic weight adjustment rule

[0054] Default weight (ecological balance scenario): biological weight = 0.5, environmental weight = 0.5;

[0055] Dynamic adjustment condition: if the number of key species (such as Siberian tigers) is detected to decrease by more than 10%, then the biological weight = 0.7, and the environmental weight = 0.3; if the environmental parameter (such as soil pH) exceeds the threshold (<6.5 or >7.5), then the environmental weight = 0.8, and the biological weight = 0.2.

[0056] Example scenario: the current number of Siberian tigers has decreased by 15% compared with last month (triggering the key species protection rule), and the weight adjustment is: biological weight = 0.7, environmental weight = 0.3.

[0057] 3. Weighted fusion calculation

[0058] Weighted biometric vector: B_weighted = biometric weight * B = 0.7 * [0.2, 0.95, 0.15, 0.5] = [0.14, 0.665, 0.105, 0.35]; Weighted environmental feature vector: E_weighted = environmental weight * E = 0.3 * [0.6, 0.65, 0.68, 0.67] = [0.18, 0.195, 0.204, 0.201].

[0059] Integrated feature vector (after concatenation): F = concat(B_weighted, E_weighted) = [0.14, 0.665, 0.105, 0.35, 0.18, 0.195, 0.204, 0.201].

[0060] The evaluation large model is a large model for evaluating ecosystem service functions. The evaluation large model is used to intelligently analyze the biometric feature vector, the environmental feature vector, the integrated feature vector, the information of animals and plants in the ecological protection zone, and the environmental quality evaluation, which can reveal the complex interaction between the biotic and environmental, and more comprehensively understand the overall situation of the ecosystem, so as to realize more comprehensive, accurate and deep ecological system service function evaluation. The evaluation large model can be optimized from the aspects of data quality, model architecture, training strategy, evaluation index, etc., to improve the accuracy and reliability of the evaluation large model for the evaluation of the ecosystem service function; the evaluation large model can also be pre-trained or transferred to a large model to pre-train on a large amount of related data to improve the performance and generalization ability of the evaluation large model, such as inputting some related data to the evaluation large model in advance to learn the evaluation large model.

[0061] The ecosystem service functions evaluated in the embodiments of the present application include, for example, biodiversity, water purification, climate regulation, soil fertility maintenance, etc. By comprehensively and deeply evaluating the ecosystem service functions, more scientific decision support can be provided for ecological protection and resource management, so as to formulate effective protection strategies.

[0062] Exemplarily, a specific example of generating the evaluation results of the ecosystem service functions is given here:

[0063] Based on the comprehensive analysis of the biotic and environmental data by the evaluation large model, the quantitative indicators of the service function modal are generated:

[0064] (1) Carbon sink (t / ha):

[0065] Input layer: Vegetation coverage data: Broadleaf forest proportion (45%), coniferous forest proportion (30%); Meteorological data: Monthly average temperature (5.2°C), historical baseline temperature (6.0°C);

[0066] Processing flow:

[0067] Vegetation carbon sink potential calculation:

[0068] Broadleaf forest carbon sink coefficient α = 0.8, coniferous forest β = 0.6;

[0069] Basic carbon sink value = (0.45 × α + 0.3 × β) = 0.54;

[0070] Temperature adaptability correction: Sigmoid response function is adopted (carbon sink efficiency should be 1.0 when the optimal temperature is met):

[0071]

[0072] Where, T: represents the current environmental temperature (5.2°C in the embodiment). T0: represents the optimal temperature of the species (or vegetation) (set to 6.0°C in the embodiment).

[0073] Ecological significance: When the environmental temperature is equal to T0, the biological function (such as carbon sink efficiency) reaches the best state.

[0074] k: represents the sensitivity coefficient (k = 0.5), which controls the steepness of the curve and reflects the sensitivity of the species to temperature deviation.

[0075] Temperature correction factor = ≈0.8.

[0076] Final carbon sink amount:

[0077] Normalized value = basic carbon sink value × temperature correction factor = 0.54 × 0.8 ≈0.43.

[0078] (2) Water conservation capacity (%)

[0079] Input layer: Hydrological data: Monthly precipitation (120mm); Soil data: pH value (7.8), snow depth (15cm); Historical baseline: Annual average precipitation (100mm);

[0080] Processing flow:

[0081] Basic conservation capacity calculation: Precipitation factor = min(current precipitation / historical baseline, 1.2) = 1.2;

[0082] Soil water holding correction:

[0083] pH deviation penalty: When the pH [6.5,7.5] when the penalty coefficient γ = 0.9;

[0084] Snow compensation: δ = h * (snow thickness / reference thickness)

[0085] Snow thickness: measured value (15 cm);

[0086] Reference thickness: take the average snow thickness in winter in this area (for example, 22.5 cm, which needs to be adjusted according to actual data);

[0087] Compensation coefficient h: take the value range 0.1~0.3 (need to be calibrated), which represents the contribution intensity of unit snow thickness to water conservation, and the value in this embodiment is 0.3;

[0088] δ = 0.3 × (15 cm / 22.5 cm) = 0.2;

[0089] Soil correction factor = γ + δ = 1.1;

[0090] Comprehensive calculation:

[0091] Original value = precipitation factor × soil correction factor = 1.2 × 1.1 = 1.32;

[0092] Normalized value = tanh (original value) ≈ 0.72.

[0093] (3) Biodiversity index

[0094] Input layer: species data: Siberian tiger index (0.58), elk index (0.15); vegetation data: broad-leaved forest (45%), shrub (25%);

[0095] Processing flow:

[0096] Species diversity calculation:

[0097] Key species weight: Siberian tiger w1 = 0.6, elk w2 = 0.4;

[0098] Animal diversity = w1 × Siberian tiger index + w2 × elk index = 0.6 × 0.58 + 0.4 × 0.15 ≈ 0.41;

[0099] Vegetation diversity calculation:

[0100] Shannon index calculation: H' = -Σ(p i ×ln(p i )), where p iPi: represents the relative proportion of the ith species (or vegetation type) in the community (i.e., the ratio of the number of individuals or coverage area of the species to the total). For example: broad-leaved forest accounts for 45% → p1= 0.45; coniferous forest accounts for 30% → p2= 0.3; shrubs account for 25% → p3= 0.25.

[0101] ln(p i ): represents the natural logarithm (base e) of p i .

[0102] If p i = 0 (a species does not exist), it is agreed that p i × ln(p i ) = 0 (to avoid mathematical undefined).

[0103] ∑ (cumulative): represents the sum of p i × ln(p i ) values for all species (or vegetation types).

[0104] Vegetation composition ratio: broad-leaved forest 0.45, coniferous forest 0.3, shrubs 0.25.

[0105] H' ≈ 1.03, after normalization 0.68.

[0106] Comprehensive index:

[0107] Geometric mean is used: sqrt(animal diversity × vegetation diversity) = sqrt(0.41 × 0.68) ≈ 0.53;

[0108] After logistic regression calibration, output 0.61;

[0109] Get service function feature vector: S = [0.65, 0.72, 0.61].

[0110] In the above implementation scheme for evaluating ecosystem service functions based on multi-dimensional data, through integration and dynamic weight adjustment of multi-dimensional data, comprehensive and dynamic evaluation of ecosystem service functions is realized, including species diversity, water purification, climate regulation, soil fertility maintenance, etc. Compared with the existing technology which usually focuses on static evaluation of single or a few service functions, the present application realizes comprehensive and dynamic evaluation of ecosystem service functions through multi-modal data fusion and dynamic analysis, which can more truly reflect the changing trend of the ecosystem.

[0111] The scheme for generating a comprehensive feature matrix is introduced below. When generating a comprehensive feature matrix based on animal and plant information, environmental quality evaluation, and ecosystem service function evaluation results, it includes:

[0112] Data preprocessing, time and space alignment of multi-source data in each spatio-temporal unit, determination of standard data, multi-source data in a single spatio-temporal unit includes comprehensive feature vector and corresponding information of animals and plants, environmental quality assessment and ecosystem service function assessment results in the current spatio-temporal unit;

[0113] Multi-modal feature extraction in standard data, generation of each modal feature vector based on the extracted features, and multi-modal features including biological features, environmental features and service function features;

[0114] In the same time and space dimension, each modal feature vector is spliced to generate a fusion feature vector of the current spatio-temporal unit;

[0115] The fusion feature vectors of multiple spatio-temporal units are arranged in time sequence and spatial grid order to form a comprehensive feature matrix.

[0116] The comprehensive feature matrix is a structured data representation generated by preprocessing, spatio-temporal alignment, feature extraction and feature fusion of multi-source heterogeneous data, and is the core input of building a digital twin model. The comprehensive feature matrix provides a unified and standardized data basis for subsequent operations.

[0117] When generating a comprehensive feature matrix based on animal and plant information, environmental quality assessment and ecosystem service function assessment results, the specific implementation process is as follows:

[0118] 1. Data preprocessing and data format unification

[0119] When generating a comprehensive feature matrix, the provided data includes animal and plant information (such as species name, location coordinates, confidence, richness index, etc.), environmental quality assessment (such as water quality parameters, soil parameters, meteorological data, etc.), ecosystem service function assessment results (such as biodiversity index, carbon sink capacity, water conservation capacity, pollination service intensity, pollutant purification rate, etc.), and also needs to combine the comprehensive feature vector obtained based on the global biological feature vector and the global environmental feature vector. The provided data is preprocessed, such as denoising, outlier removal, missing value filling, data format unification (such as rasterization, time series alignment), and data processing based on dimensionless normalization or standardization.

[0120] 2. Spatio-temporal alignment and gridding

[0121] The data is time and space aligned based on the unified space-time reference to determine the standard data. For example, a unified spatial grid (e.g. 10m x 10m) is divided in the spatial dimension and a unified time resolution (e.g. 1 hour, 5 hours, etc.) is determined in the time dimension; when aligning, for example, the following methods are used: using spatial interpolation to map species observation point data to the grid, using time interpolation to align data of different sampling frequencies to a unified timestamp.

[0122] 3. Feature extraction and vectorization

[0123] Multi-modal feature extraction is performed in the standard data, where the multi-modal features at least include biological features, environmental features, and service function features. Biological features include image features, audio features, and positioning features. For image data, image features are extracted using a convolutional neural network, for audio data, voiceprint features are extracted using an audio recognition model, and for positioning data, positioning features are extracted using a graph neural network or trajectory embedding model. Environmental data at least includes meteorological data and physical environment data, and time series statistical features are extracted for environmental data to obtain environmental features. Service function data is used to extract service function features. After extracting features to obtain fixed-dimensional feature vectors, all feature vectors are organized by spatial grid-time unit. It is worth noting that in the same time and space dimension, the weights of each modality are preferably dynamically adjusted to obtain weighted feature vectors of each modality with dynamically adjusted weights.

[0124] 4. Feature fusion and matrix generation

[0125] In feature fusion, feature vectors of different modalities are spliced in the same time and space dimension, and feature importance can also be considered to assign different weights to feature vectors of different modalities. After feature fusion, each spatial grid-time unit corresponds to a fused feature vector, and all fused feature vectors are arranged in time sequence and spatial grid order to form a comprehensive feature matrix, which is used as input for subsequent modeling.

[0126] 4.1 Assume input data (single space-time unit)

[0127] Space-time unit: Grid A of an ecological protection area, time: 2023-10-01 10:00.

[0128] The input data includes the following content:

[0129] Animal and plant information (biological features): Species 1 (Northeast Tiger): Quantity = 2, Confidence = 0.95, Activity Range = 5 km²; Species 2 (Elk): Quantity = 15, Confidence = 0.90.

[0130] Environmental quality assessment (environmental features): Temperature = 18°C, Humidity = 65%, Soil pH = 6.8, Snow depth = 20 cm.

[0131] Ecosystem service function assessment results (service function features): Carbon sink capacity = 2.4 t / ha, Water conservation capacity = 85%, Biodiversity index = 0.78.

[0132] 4.2 Generate fusion feature vector for single spatio-temporal unit

[0133] (1) Data preprocessing

[0134] Numerical normalization (scaled to [0, 1]):

[0135] Northeast tiger quantity: 2 → normalized 0.2 (assuming the maximum number of tigers in this area = 10); Temperature: 18°C → 0.6 (assuming temperature range [-10, 30] °C); Carbon sink capacity: 2.4 t / ha → 0.8 (assuming maximum carbon sink capacity = 3 t / ha).

[0136] (2) Multi-modal feature extraction

[0137] Biological feature vector (dimension = 4): [0.2, 0.95, 0.15, 0.5] (Note: 0.15 = normalized value of elk quantity, 0.5 = normalized value of activity range); Environmental feature vector (dimension = 4): [0.6, 0.65, 0.68, 0.67] (Note: 0.65 = normalized value of humidity, 0.68 = normalized value of soil pH, 0.67 = normalized value of snow depth); Feature vector of service function modality (dimension = 3): [0.8, 0.85, 0.78] (Note: 0.85 = normalized value of water conservation capacity).

[0138] It should be noted that in this embodiment, dynamic adjustment of the weights of each modality is still preferably considered.

[0139] The weights of the three types of vectors need to satisfy the normalization condition: biological weight + environmental weight + service weight = 1; The following gives an example of dynamic weight calculation: basic weight (default scenario): biological weight = 0.4, environmental weight = 0.4, service weight = 0.2.

[0140] Exemplary adjustment rules: If the species diversity index decreases by more than a threshold value (such as 10%), then: biological weight = 0.6, environmental weight = 0.3, service weight = 0.1.

[0141] If the water quality pH exceeds the safe range (6.5-7.5), then: environmental weight = 0.7, biological weight = 0.2, service weight = 0.1.

[0142] In the scenario of key species protection priority, input data: Northeast Tiger population month-on-month decrease by 15% (trigger protection rule); environmental parameters normal (pH = 6.8, temperature = 18°C); carbon sink stable, weight adjustment: biological weight = 0.6, environmental weight = 0.3, service weight = 0.1.

[0143] Weighted feature vector:

[0144] Biological feature vector B = [0.2, 0.95, 0.15, 0.5]→0.6*B = [0.12, 0.57, 0.09, 0.3];

[0145] Environmental feature vector E = [0.6, 0.65, 0.68, 0.67]→0.3*E = [0.18, 0.195, 0.204, 0.201];

[0146] Service function vector S = [0.8, 0.85, 0.78]→0.1*S = [0.08, 0.085, 0.078].

[0147] 4.3 Feature Splicing

[0148] Splice the vectors of the three modalities into a fusion feature vector (dimension = 11):

[0149] Fusion feature vector: F = concat(0.6B, 0.3E, 0.1S) = [0.12, 0.57, 0.09, 0.3, 0.18, 0.195, 0.204, 0.201, 0.08, 0.085, 0.078].

[0150] 4.4 Constructing Comprehensive Feature Matrix

[0151] Repeat the above process to generate fusion feature vectors for multiple spatio-temporal units and arrange them in spatio-temporal order:

[0152]

[0153] The final comprehensive feature matrix is in the form of, for example:

[0154] [0.12, 0.57, 0.09, 0.30, 0.18, 0.195, 0.204, 0.201, 0.08, 0.085, 0.078], Grid A 10:00;

[0155] [0.18, 0.552, 0.108, 0.30, 0.186, 0.189, 0.210, 0.195, 0.079, 0.083,0.077], Grid B 11:00;

[0156] [0.00, 0.00, 0.00, 0.00, 0.165, 0.210, 0.195, 0.180, 0.07, 0.075,0.065], Grid C 10:00.

[0157] And in the above implementation process, consistency check can be carried out to ensure the spatial consistency of different data sources in the same grid; integrity check can be carried out to identify whether there are null values or abnormal values; and logical check can also be carried out to verify the ecological rationality between species distribution and environmental conditions (such as wetland species should not appear in arid areas).

[0158] In the above implementation scheme of generating the comprehensive feature matrix, the framework of "species-environment-service function" three-modal feature fusion is realized, which aligns the three types of data of species, environment and service function to the unified grid and time axis, eliminates the scale difference, and ensures the comparability of variables in the same dimension; the information of animals and plants supplements the biological level data, the environmental data describes the background information of species, and the service function provides the ecological output index, which are complementary to improve the information quantity; through the spatio-temporal alignment and deep learning, the quantization of the coupling relationship of ecological variables is realized, the matrix data retains the effective features, and the data signal-to-noise ratio is improved.

[0159] In the embodiment of the present application, after generating the comprehensive feature matrix, a digital twin engine is constructed by deep learning on the comprehensive feature matrix, which includes the following steps:

[0160] Deep learning is performed on the comprehensive feature matrix to output a weight matrix, a response function and a threshold rule, the weight matrix includes multiple coupling weights and is used to quantify the action relationship between species and environment, service function, the response function is used to describe the response relationship between ecological system variables, and the threshold rule is used to define the critical state of the ecological system;

[0161] Based on the weight matrix, the response function and the threshold rule, a dynamic kernel for calculating the state of the ecological system, evaluating the trend of the state change of the ecological system and identifying the critical state is generated;

[0162] The dynamic kernel and the rule engine are packaged to construct a digital twin engine, and the rule engine makes decision response based on the output results of the dynamic kernel.

[0163] After generating the comprehensive feature matrix, the comprehensive feature matrix is subjected to deep learning by a deep learning model (such as a graph neural network), and the coupling weights between species-environment, species-service function, environment-service function, species and environment-service function are automatically extracted, and a weight matrix, a response function and a threshold rule are output, which together constitute a dynamic kernel to provide logical support for subsequent calculations. Then, taking the dynamic kernel as the calculation core, a rule engine is integrated and encapsulated into a digital twin engine. The digital twin engine has the capabilities of real-time data access, state calculation, trend analysis, rule judgment and response triggering, and is the core driving module of the digital twin large model, supporting the dynamic operation and intelligent decision-making of the large model.

[0164] Among them, the dynamic kernel is responsible for real-time calculation of the interaction and state change between each element (species, environment, service function) of the ecosystem, analysis of the state change trend of the ecosystem, identification of critical situations, and output of the current system state, analyzed change trend, and identification of whether the critical condition is reached; the rule engine performs logical judgment and response actions based on the output of the dynamic kernel to realize intelligent feedback and decision support.

[0165] Specifically, for the dynamic kernel, it is mainly used for calculating the real-time state of the current ecosystem, analyzing the state change trend of the ecosystem and identifying the critical state. For example, in calculating the real-time state of the ecosystem, the state of the species (species richness, diversity, distribution, etc.), the state of the environment (water quality, soil, weather, etc. environmental parameters), and the state of the service function (carbon sequestration capacity, water conservation capacity, biodiversity maintenance capacity, etc.) are evaluated. In analyzing the state change trend of the ecosystem, the latest data, historical data of the ecosystem and the response function are used to calculate the change direction and amplitude of the current state relative to the previous time; for example, the upward or downward trend of species richness, the improvement or deterioration trend of water quality, and the strengthening or weakening trend of service capacity. In identifying the critical state, whether the ecological threshold (such as whether the pH is less than 6.5 or the species richness is decreased by more than 20%) is reached is identified. The dynamic kernel can provide input content for the rule engine, such as passing the state value, state change trend and critical identification result of the current ecosystem to the rule engine, which will process accordingly based on the input content.

[0166] It should be noted that the dynamic kernel mainly processes real-time data, but the historical data of the ecosystem as background information enhances the accuracy and reliability of the analysis and is used to assist in a more comprehensive and accurate understanding of the changes in the ecosystem.

[0167] For the rule engine, it is mainly responsible for executing the following tasks based on the current state, state change trend and critical identification results output by the dynamic kernel: 1. Logical judgment and decision triggering; 2. Response action execution. When logical judgment and decision triggering, it is judged whether the preset business rules or response conditions are met, such as triggering acidification warning when pH is less than 6.5, triggering ecological degradation warning when species richness decreases by more than 20%, and triggering service adjustment suggestion when carbon sink valuation decreases by more than 10%. When executing response actions, based on the judgment results, matching actions are automatically executed, such as issuing warning information and recommending intervention measures (such as planting water purification plants and limiting pollution, etc.). The rule engine can also support feeding back the judgment results and response actions to the visualization interface for relevant personnel to view risks and intervention suggestions, providing auxiliary information for ecological protection and resource regulation decision-making. It should be noted that the rule engine not only considers the current state, but also combines the state change trend for logical judgment, so as to realize more intelligent and forward-looking ecological response and decision support.

[0168] That is, the dynamic kernel calculates the current state of the ecological system, analyzes the state change trend and critical identification, and provides the basis for real-time monitoring and response; the rule engine, based on the current state, state change trend and critical identification results output by the dynamic kernel, combines the preset business rules, executes logical judgment and triggers corresponding response actions, realizes intelligent feedback and decision support of the ecological system. As the key module to realize intelligent feedback and decision support, the rule engine forms a "judgment + response" closed loop mechanism with the dynamic kernel, and builds a digital twin engine as the core driving module of the digital twin big model.

[0169] Optionally, when the comprehensive feature matrix is subjected to deep learning, the weight matrix, response function and threshold rule are output, including the following steps:

[0170] The comprehensive feature matrix is input into the deep learning model to learn the coupling relationship between species variables, environmental variables and service function variables, and the weight matrix including the coupling weight between species and environment, the coupling weight between species and service function, the coupling weight between environment and service function, and the coupling weight of multi-element synergy is output;

[0171] A nonlinear activation layer is embedded in the deep learning model to fit the functional relationship between species variables, environmental variables and service function variables, and determine the response function;

[0172] During the training process of the deep learning model, the key ecological threshold is learned based on the loss function constraint, and the threshold rule is generated.

[0173] In the deep learning model, the integrated feature matrix is inputted, and the coupling weights of species and environment, the coupling weights of species and service function, the coupling weights of environment and service function, and the coupling weights of multi-element synergy are obtained by deep learning on the integrated feature matrix to form a weight matrix. The coupling weights of multi-element synergy can be understood as the joint influence or coupling weights of species and environment on ecosystem service function. It not only considers the influence of species on service function or the influence of environment on service function, but also comprehensively considers the interaction between species and environment and the synergistic effect of them on service function (such as carbon sink, water conservation, biodiversity maintenance, etc.). As an example, the species-environment-service function synergy weight indicates how the change of species responds to the service function under specific environmental conditions, or indicates how the environmental change affects the service function under specific species composition. For example, under the joint action of increased precipitation (environmental factor) and improved vegetation coverage (species factor), the water conservation capacity is significantly improved, and the coupling weights of multi-element synergy represent the positive influence strength of the joint action of the two on water conservation. For another example, under the joint action of soil acidification (environmental factor) and reduction of key species (species factor), the soil carbon sink capacity is significantly reduced, and the coupling weights of multi-element synergy represent the negative influence strength of the joint action of the two on carbon sink.

[0174] The species, environment and service function are in the form of variables in the deep learning model, for example, the species variables include species richness, diversity index, presence or absence of key species, the environmental variables include water quality pH, soil organic matter, temperature, precipitation, and the service function variables include carbon sink amount, water conservation capacity, pollination service intensity, etc. Learning the coupling relationship between species, environment and service function is essentially learning the coupling relationship between variables, and learning the coupling relationship between species, environment and service function is essentially modeling the interaction and response mechanism between variables through deep learning. The deep learning model automatically extracts the coupling weights, response functions and threshold rules between them by learning the correlation, causality or response mode between these variables.

[0175] The response function represents how one variable (such as an environmental factor) changes when another variable (such as species richness or service function) changes, and is used to describe the response relationship between ecosystem variables; it defines the influence mode and strength between variables, which can be linear, nonlinear or conditional relationship. For example, the response function can represent the improvement of water quality on the improvement of species richness, or the increase of vegetation coverage on the enhancement of carbon sink capacity.

[0176] The implementation process of deep learning on the integrated feature matrix to output the weight matrix, the response function and the threshold rule is briefly described as follows:

[0177] The comprehensive feature matrix is input into the deep learning model, the coupling relationship between the species, the environment and the service function is learned through model training, and a coupling weight matrix between the species, the environment and the service function is output; a nonlinear activation layer is embedded in the deep learning model to learn the response relationship between variables, and a response function is obtained; a loss function constraint is introduced in the model training process to automatically learn the key ecological threshold, and a threshold rule is obtained.

[0178] The deep learning model is, for example, a graph neural network (GNN), a spatio-temporal Transformer, a convolutional neural network (CNN) or the like, which is used to learn the coupling relationship between the species, the environment and the service function from the comprehensive feature matrix; the nonlinear activation layer is used to introduce a nonlinear transformation, so that the model can fit a more complex function relationship; the nonlinear activation layer is embedded in the model to enhance the modeling capability of the model for the nonlinear response relationship between the variables in the ecosystem, and to fit the complex function relationship between the species, the environment and the service function variables in the ecosystem, so as to determine the response function; by introducing the loss function constraint, the threshold information can be learned, and then the threshold rule is obtained.

[0179] In the above embodiment of constructing the digital twin engine, the dynamic kernel for calculating the state of the ecosystem, evaluating the trend of the state change and identifying the critical state is generated by deep learning on the comprehensive feature matrix, and the digital twin engine is constructed by integrating the dynamic kernel and the rule engine, which can provide guarantee for the generation of the subsequent digital twin large model.

[0180] The process of initializing the digital twin large model is introduced below. In the case of constructing the digital twin engine based on deep learning, the model static base information used to construct the three-dimensional spatial skeleton and the basic attributes is loaded to the digital twin engine; the digital twin engine and the model static base information are spatially registered and logically bound, and various model parameters in the digital twin engine are configured to initialize the digital twin large model.

[0181] The model static base information is used to construct the three-dimensional spatial skeleton and the basic attributes of the digital twin large model. The model static base information provides the terrain structure and the ecological attributes to construct the three-dimensional spatial skeleton and the basic attributes of the digital twin large model, which is the spatial and attribute basis for the operation of the large model.

[0182] The three-dimensional spatial skeleton, for example, includes terrain DEM (Digital Elevation Model), spatial gridding, and spatial positioning. The terrain DEM provides the elevation, slope, slope direction, and other information of each spatial point, thereby providing continuous elevation and terrain attributes; spatial gridding is to divide the ecological region into regular grids (such as 10 m x 10 m), each grid corresponds to a spatial unit, and then the DEM is discretized into numbered grid units; spatial positioning is to position and superimpose all dynamic data (species distribution, environmental parameters) based on the skeleton, that is, to accurately map the real-time data of species, environment, etc. to the corresponding terrain unit using grid ID.

[0183] The basic attributes, for example, include vegetation type grid and soil texture map, the vegetation type grid provides vegetation gridded attribute values, and the soil texture map provides soil gridded attribute values; by unifying the grid coordinates, the vegetation, soil, and dynamic data are aligned under the same spatial reference, avoiding misalignment. For the vegetation type grid, by marking specific vegetation types (forest, grassland, wetland, etc.) in each spatial grid, the coverage characteristics are described, and these vegetation types directly affect the habitat suitability, carbon sink capacity, and other ecological processes of species, thereby providing key ecological attribute inputs for the digital twin macro model. The soil texture map provides soil ecological parameters by giving the soil type, organic matter content, pH value, and other attributes of each grid, which have important influences on plant growth rate, nutrient supply, pollutant adsorption, and water purification efficiency, and are the basic inputs for the digital twin macro model to simulate ecological processes.

[0184] The process of initializing the digital twin macro model based on the digital twin engine and the model static base information mainly includes the following steps: 1. loading the model static base information into the digital twin engine to provide the three-dimensional spatial skeleton and the basic attributes for the macro model, ensuring the accuracy and authenticity of the macro model in space; 2. integrating the model static base information with the dynamic kernel and rule engine in the digital twin engine through spatial registration and data association, establishing the corresponding relationship between space and attributes, so that the digital twin engine can perform state calculation and response judgment based on the model static base information; 3. configuring the parameters in the digital twin engine, such as coupling weight, response function, threshold rule, etc., according to the characteristics of the actual ecological system (based on historical long-term statistics, such as multi-year average species richness, seasonal water quality fluctuation range, soil background value, etc.), and initializing the state of the macro model, ensuring that the macro model can accurately reflect the initial situation of the ecological system.

[0185] The coupling weight is used to quantify the interaction strength between species and environment, service function, is a key parameter in the large model for describing the relationship between variables; the response function is used to describe the response relationship between variables, is a function for describing the change rule between variables in the large model; the threshold rule is used to define the critical state of the ecosystem, is used to identify whether the warning or response condition is reached, is the basis for logical judgment and response triggering in the large model. These parameters together constitute the dynamic kernel of the digital twin large model, are the key model parameters in the digital twin large model for describing the relationship between variables, change rule and critical state of the ecosystem, are the core components of the digital twin large model that can accurately reflect the state and behavior of the ecosystem, and are also the basis for the large model to realize real-time calculation, state update, trend analysis, critical identification and response triggering.

[0186] After initializing the digital twin large model, real-time data processing is performed based on the digital twin large model, and the linkage state of species, environment and ecosystem service function is rendered on the visualization interface. The process includes the following steps:

[0187] Based on the data interface, the obtained real-time data is accessed to the digital twin engine, and the digital twin engine updates the model state of the digital twin large model. The real-time data includes real-time plant and animal information, real-time environmental quality assessment, and real-time ecosystem service function assessment results;

[0188] Based on the updated digital twin large model, the real-time data and the historical data of the ecosystem are processed, and the ecosystem operation result is output. The ecosystem operation result indicates the linkage state of species, environment and ecosystem service function, and the ecosystem operation result includes at least one of the current state of the ecosystem, the change trend of the state of the ecosystem, the critical state of the ecosystem, and the decision response;

[0189] The ecosystem operation result provided by the digital twin large model is visualized through a VR interface or an AR interface.

[0190] The digital twin engine obtains real-time data (real-time plant and animal information, real-time environmental quality assessment, and real-time ecosystem service function assessment results) through a data interface (such as an Internet of Things interface). The digital twin engine dynamically updates the state of the digital twin large model based on the obtained real-time data, and keeps the large model synchronized with the real ecosystem.

[0191] After updating the digital twin macro model, real-time data and ecosystem historical data are processed based on the updated digital twin macro model. During the processing, the digital twin engine uses a dynamic kernel to perform state calculation based on real-time data, ecosystem historical data, and model static base information, assesses the current state of the ecosystem, the state change trend, and determines whether the critical state is reached; the rule engine performs logical judgment and response actions according to the output of the dynamic kernel, such as triggering an early warning, to provide decision support for ecological protection and management.

[0192] Processing real-time data and ecosystem historical data using the digital twin macro model can output an ecosystem operation result indicating the linkage state of species, environment, and ecosystem service functions, and the ecosystem operation result includes at least one of the current state of the ecosystem, the state change trend of the ecosystem, the critical state of the ecosystem, and the decision response. The ecosystem operation result provided by the digital twin macro model is displayed through a visual interface to display the latest data based on a VR interface or an AR interface.

[0193] It should be noted that the ecosystem historical data reveals the long-term evolution law and seasonal change of the ecosystem variables, which can help the macro model understand the formation process of the current state; and in combination with real-time data and historical data, the macro model can capture the long-term trend and short-term fluctuation of the ecosystem variables, thereby improving the accuracy of the analysis. That is, the ecosystem historical data is an important input for the analysis of the digital twin macro model, which not only provides background information and long-term trends, but also enhances the accuracy of the analysis.

[0194] Figure 2 A simplified flowchart of the digital twin macro model processing real-time data provided by the embodiments of the present application is shown.

[0195] The dynamic kernel receives real-time data accessed through the Internet of Things (real-time plant and animal information, real-time environmental quality evaluation, and real-time ecosystem service function evaluation results), combines ecosystem historical data, model static base information, coupling weights, response functions, and threshold rules, calculates the current state of the ecosystem, assesses the state change trend of the ecosystem, identifies whether the state of the ecosystem reaches a critical condition, and outputs an output result including the current state value of the ecosystem, the state change trend, and the critical identification condition.

[0196] The rule engine performs business logic judgment and response actions based on the output of the dynamic kernel. The rule engine receives the current state value of the ecosystem, the state change trend, and the critical identification condition output by the dynamic kernel, judges whether the early warning condition is met based on the set response strategy, and performs response and outputs decision suggestions.

[0197] The decision response provided by the rule engine, and the current state value of the ecosystem output by the dynamic kernel, the state change trend, and the critical identification situation are rendered on the visual interface to refresh the large model in real time through the real-time data driven digital twin engine of the digital twin large model connected by the Internet of Things, to ensure that the large model is synchronized with the real ecosystem.

[0198] The ecosystem operation results provided by the digital twin large model include the current state of the ecosystem, the state change trend, the critical state identification, and the response action, comprehensively reflecting the operation status and response mechanism of the ecosystem.

[0199] The current state of the ecosystem includes the state of species, the state of environmental quality, and the state of ecosystem service function. The state of species includes, for example, the real-time identified species name, richness, diversity index, and spatial distribution heat map in each grid. The state of environmental quality includes, for example, the real-time grid values of water quality, soil, and weather. The state of ecosystem service function includes, for example, carbon sink flux, water conservation capacity, biodiversity maintenance index, and pollutant purification efficiency in grid form. The real-time environmental quality assessment is a comprehensive evaluation based on raw data, and the digital twin large model outputs continuous physical quantities (such as pH, dissolved oxygen, and organic matter) for each grid after rasterizing the evaluation results, i.e., obtains the environmental state. The real-time ecosystem service function evaluation result is a comprehensive scoring or grading evaluation of raw data, and the digital twin large model converts the evaluation results into continuous physical quantities or fluxes for each grid after rasterizing, i.e., service functions.

[0200] When the current state of the ecosystem is displayed on the visual interface, the species distribution map, the environmental quality map, and the service function state map can be displayed. The specific forms of the species distribution map include, for example, a heat map (using color depth to represent species density or richness), a point map (using discrete points to mark the location of each observed individual or sampling point), and a three-dimensional model (superimposing species data on a three-dimensional terrain or vegetation model to provide an immersive spatial perspective). The three forms can be presented individually or in combination to form a complete visualization of the spatial distribution of species. The environmental quality map, for example, plots parameters such as water quality (pH, dissolved oxygen), soil (organic matter, heavy metals), and weather (temperature, humidity) into a spatial distribution map by grid or region, such as a pH distribution map or a temperature heat map, to quickly identify where the water quality is acidic and where the temperature is high. The service function state map, for example, plots the output capacity of the ecosystem (carbon sink capacity, water conservation capacity, biodiversity index, etc.) into a spatial distribution map to visually display carbon sink high-value areas, water conservation hotspots, and biodiversity cold and hot spots.

[0201] The change trend includes the change trend (such as rising, falling, fluctuation) of each element (species, environment, service), the change rate and amplitude (such as the daily change amount of the water quality index). The state change trend can be embodied by a trend curve, a dynamic evolution diagram, and a change rate diagram. For example, the trend curve indicates the change curve of the species richness, the water quality index, and the service index over time; the dynamic evolution diagram intuitively presents the change process of the ecosystem state over time in the form of animation or time axis playback; and the change rate diagram shows the rising and falling speed of each ecological element in unit time in the form of spatial distribution.

[0202] The critical state identification includes identifying whether the ecological threshold is reached. The pre-warning map can be used to display which area triggers the ecological threshold pre-warning, the pre-warning list can be used to list all the current pre-warning information (such as pH being too low and species richness decreasing), and the critical identification can be used to highlight the critical area.

[0203] For response actions and decision suggestions, the large model can automatically trigger response actions and recommend intervention measures (such as planting water purification plants, limiting pollution discharge, and ecological restoration schemes), and can use scenario simulation to simulate the ecological restoration effect under different intervention measures.

[0204] The above implementation scheme is based on real-time data to refresh the large model, ensuring that the large model is synchronized with the real ecosystem; the running results output by the digital twin large model cover the current state of the ecosystem, the state change trend, the critical pre-warning, and the response action, comprehensively reflecting the running condition and response mechanism of the ecosystem; and the visual interface displays all the information that needs to be intuitively understood, interacted, or decision-referenced by the user, providing a good experience for the user.

[0205] In an optional embodiment of the present application, the method further comprises:

[0206] In response to receiving the interactive operation on the VR interface or the AR interface, updating the model state of the digital twin large model, and obtaining the updated ecosystem running result output by the digital twin large model;

[0207] Based on the updated ecosystem running result, updating the visual content on the VR interface or the AR interface in real time.

[0208] The running result of the digital twin large model is visualized by VR or AR technology, which enables the user to intuitively understand the current state, state change trend, critical situation, and response decision of the ecosystem. The visual interface supports interactive operation, the model state of the digital twin large model can be updated based on the interactive operation of the user on the visual interface, the latest ecosystem running result is output by the digital twin large model, and the visual interface is updated.

[0209] Visual interfaces can support user interaction in ways such as the following:

[0210] Parameter Adjustment: Allows users to adjust parameters such as environmental factors and species numbers, with the large model providing the response. Scenario Simulation: Offers different scenario options; users can start simulations and view changes in the simulated ecosystem under different scenarios. Layer Control: Users can choose to show or hide different data layers, such as species distribution and water quality. Warning Information Viewing: Displays warning details; users can click to view specific risk areas and recommended measures. Timeline Operation: Users can drag the timeline to view historical changes and trends in the ecosystem's state.

[0211] The aforementioned supported interaction methods enable users to actively participate in model operations, explore different scenarios, and verify results, thereby gaining a better understanding of the ecosystem's state and aiding in decision-making.

[0212] The visualization interface can be either a VR or AR interface. The VR interface supports immersive operation; for example, users can wear VR headsets to enter a three-dimensional virtual ecosystem and interact with the virtual environment through controllers, gestures, and eye tracking to view the ecological status of different areas and observe species distribution, water quality changes, etc. In parameter adjustment and scenario simulation scenarios, environmental parameters (such as precipitation and temperature) can be adjusted in the virtual interface, and different scenario simulations (such as pollution events and ecological restoration) can be initiated to view the response in real time. In information query and feedback scenarios, clicking on a species or area allows users to view detailed information (such as species name, richness, and health status).

[0213] The AR interface supports augmented reality overlay, allowing virtual ecological information to be superimposed onto real-world scenes using AR glasses or a camera. This means that when users wear AR glasses or use their phone's camera, they can see real-time monitoring data overlaid on the real-world environment, such as water quality indicators displayed above a real river or species distribution maps overlaid on forest land. Users can view real-time monitoring data (such as water and air quality) in the real environment and can click to view details, historical trends, and early warning information. In scenario simulations, the effects of different intervention measures (such as vegetation restoration and pollution control) can be simulated, and the differences between the current ecological state and the simulated state can be compared. Users can operate the interface via voice commands or gesture input.

[0214] In this embodiment, the operation of the large model relies on a real-time update mechanism (IoT data-driven) and visualization (VR / AR interactive interface) to realize real-time mapping and dynamic interaction of the ecosystem. The VR / AR interface supports users to interact intuitively and in real-time with the digital twin large model through immersive interaction, parameter adjustment, scenario simulation, information query, etc., thereby improving the efficiency of understanding, analysis and decision-making of the ecosystem.

[0215] Figure 3 The overall implementation flowchart of the digital twin body construction of the embodiment of the application is shown.

[0216] Step 301, generating a biological feature vector based on biological data collected in the ecological protection zone, generating an environmental feature vector based on environmental data collected in the ecological protection zone, performing species identification on the biological feature vector based on a biological large model to obtain information of animals and plants in the ecological protection zone, and performing environmental analysis on the environmental feature vector based on an environmental large model to obtain an environmental quality evaluation situation in the ecological protection zone.

[0217] Step 302, splicing the biological feature vector and the environmental feature vector in combination with dynamically adjusted biological weights and environmental weights to determine a comprehensive feature vector.

[0218] Step 303, performing comprehensive reasoning analysis on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the information of animals and plants, and the environmental quality evaluation situation based on an optimized evaluation large model to generate an ecosystem service function evaluation result.

[0219] Step 304, performing data preprocessing and time and space alignment of data on multi-source data in each space-time unit to determine standard data, and the multi-source data in a single space-time unit includes information of animals and plants corresponding to the current space-time unit, an environmental quality evaluation situation, and an ecosystem service function evaluation result.

[0220] Step 305, performing multi-modal feature extraction in the standard data, generating a feature vector of each modality based on the extracted features, obtaining a weighted feature vector of each modality in the same time and space dimensions, and obtaining a dynamically adjusted weight of each modality; splicing the weighted feature vectors of each modality to generate a fusion feature vector of the current space-time unit, arranging the fusion feature vectors of multiple space-time units in a time sequence and a space grid order to form a comprehensive feature matrix, and the multi-modal features at least include biological features, environmental features, and service function features.

[0221] Step 306, performing deep learning on the comprehensive feature matrix to output a weight matrix, a response function, and a threshold rule, and generating a dynamic kernel for calculating an ecosystem state, evaluating a state change trend, and identifying a critical state based on the weight matrix, the response function, and the threshold rule.

[0222] Step 307, packaging the dynamic kernel and the rule engine to construct a digital twin body engine.

[0223] Step 308, load the model static base information for constructing the three-dimensional space skeleton and the basic attribute to the digital twin engine, spatially register and logically bind the digital twin engine and the model static base information, and configure various model parameters in the digital twin engine to initialize the digital twin large model.

[0224] Step 309, in response to inputting real-time data of the digital twin large model, updating the model state of the digital twin large model, processing real-time data and ecosystem historical data based on the updated digital twin large model, outputting ecosystem running results and displaying through a visual interface.

[0225] Step 310, in response to receiving an interactive operation on the visual interface, obtaining the latest ecosystem running results and updating the visual content on the visual interface.

[0226] The above implementation process of the present application realizes real-time twinning of the ecosystem state based on digital twin technology; the running results output by the digital twin large model cover multiple dimensions, comprehensively reflecting the running state and response mechanism of the ecosystem; and the use of the visual interface for content display and supporting real-time interaction provides a good experience for users.

[0227] In order to further introduce the scheme provided by the embodiments of the present application, a specific example is used for illustration.

[0228] Biological data and environmental data of the ecological protection zone are collected. When collecting biological data, the following operations are performed: multiple infrared cameras are deployed in the hot spot area of the Siberian tiger, and a group of images are taken every interval (such as 15 minutes), which are transmitted back to the edge computing node through the wireless network; for example, 10 Siberian tigers are worn with GPS collars (accuracy ± 3m), and the position data is uploaded to the Beidou satellite every 30 minutes; for example, 20 point positions are deployed with acoustic sensors to record tiger howling audio. When collecting environmental data, the following operations are performed: install a weather station to monitor snow depth, temperature, humidity, and generate a forest three-dimensional point cloud every month by unmanned aerial vehicle scanning.

[0229] After collecting data, data preprocessing is performed, such as time and space alignment of data. When time alignment, the GPS collar data and weather data are converted into a unified format; the time deviation of audio data and image data is calibrated by dynamic time warping algorithm. When spatial alignment, the point cloud data and GPS coordinate projection are converted to UTM.

[0230] After data preprocessing, the generation of feature vectors is carried out. In the generation of biological feature vectors, for example, 2048-dimensional feature vectors are extracted from image data, and for example, 128-dimensional embedding vectors are output from audio data, and GPS position information is implicitly aligned in the spatial alignment step, and no further encoding is repeated in the feature stage. In the generation of environmental feature vectors, the numerical values of snow depth, temperature, etc. are normalized and spliced into a 6-dimensional vector. The biological feature vectors and environmental feature vectors are weighted and fused using an attention mechanism to generate a comprehensive feature vector (dimension: 2048+128+6=2182), which can dynamically adjust the biological weight and environmental weight, such as based on the real-time contribution of biological factors and environmental factors to dynamically adjust the biological weight and environmental weight, to more accurately reflect the current state of the ecosystem.

[0231] After obtaining the biological feature vector, the environmental feature vector, and the comprehensive feature vector, the biological feature vector is processed based on the biological large model to identify the species, the environmental feature vector is processed based on the environmental large model to predict the snow depth sequence in the next 7 days (such as [28, 26, 25, 24, 23, 22, 21] cm), and the biological feature vector, the environmental feature vector, the comprehensive feature vector, the identification result of the biological large model, and the output result of the environmental large model are processed based on the evaluation large model to evaluate the ecosystem service function.

[0232] Based on the identification result of the biological large model, the output result of the environmental large model, and the evaluation result of the ecosystem service function, a comprehensive feature matrix is constructed. The shape of the comprehensive feature matrix is, for example, N x 2182, where N is the time step (such as daily data for the past 30 days), and each row contains 2182-dimensional features (biological + environmental + service function evaluation features). A dynamic kernel is developed by deep learning the comprehensive feature matrix, which includes the following processes: weight matrix learning, response function fitting, and threshold rule generation.

[0233] In the weight matrix learning stage, a suitable model architecture is selected, position encoding is added to the comprehensive feature matrix to preserve the time sequence information, and a 2182 x 2182 weight matrix is calculated by self-attention, representing the coupling strength between species-environment-service function. Non-zero elements in the weight matrix represent key relationships (such as the weight of Northeast Tiger activity and snow is -0.6).

[0234] In the response function fitting stage, the key coupling relationships in the weight matrix (such as the weight of Northeast Tiger activity and snow -0.6), and the historical time series data (Northeast Tiger activity range, snow depth, deer density) are used as input data to describe the influence of snow and deer changes on the Northeast Tiger activity range, and the fitting accuracy and readability are considered to obtain the response function.

[0235] In the threshold rule generation stage, the historical northeast tiger activity change rate, snow thickness, and deer group frequency are set according to statistical rules to set a normal interval; if the change rate is far below the average for 5 consecutive days and the snow is very thin, the habitat is marked as degraded, and the threshold is refreshed every month according to the latest data to adapt to seasonal fluctuations.

[0236] After developing the dynamic kernel, based on the dynamic kernel and the rule engine, the digital twin engine is constructed, and the rule engine makes decision response based on the output result of the dynamic kernel. For example, in the case of 28 cm of snow falling for 4 days and the northeast tiger activity range decreasing by 18%, the dynamic kernel determines that it exceeds the -15% warning line, and the rule engine immediately triggers the "habitat crisis" alarm.

[0237] In the case of constructing a digital twin large model based on the digital twin engine and the model static base information, real-time data is accessed to the digital twin engine based on the data interface, and the model state of the digital twin large model is updated, and the digital twin large model processes real-time data and ecosystem historical data to provide ecosystem running results for display on a visualization interface (such as an AR interface).

[0238] The AR interface display content includes, for example: a northeast tiger individual, a snow heat map, a deer group density particle cloud, a key indicator card, an event warning area, etc. Each northeast tiger is represented by a 3D model similar to its real body shape, the 3D model is anchored by real-time GPS coordinates, and moves synchronously with the animal, a trajectory line is displayed behind the 3D model showing the moving path in the past few hours or days, and a number can be suspended above the 3D model, which can indicate gender, age, and other identity information. In the snow heat map, the thinner the snow, the bluer the color, and the thicker the snow, the whiter the color. The color immediately recalculates and synchronously refreshes every time the thickness changes, and the snow surface color seen by the human eye changes in real time. For the deer group density particle cloud, it is a "cloud" composed of countless small light points appearing in the air above the forest. The brighter and denser the light points, the more deer groups there are in the area; sparse or dim light points indicate fewer deer groups. Users can control the appearance or hiding of the particle cloud through voice commands or related gestures. The key indicator card can display habitat score, warning threshold, and other information, allowing managers to quickly understand relevant content. For the event warning area, after determining that the "habitat crisis" rule is triggered, the affected area will be immediately covered with a red semi-transparent grid, and a prompt box will pop up, telling the ranger to prioritize patrol and giving route guidance.

[0239] The embodiments of the present application also provide a digital twin construction system for a terrestrial ecosystem, as shown in Figure 4 , comprising:

[0240] The acquisition module 401 is configured to acquire animal and plant information and environment quality evaluation information in the ecological protection zone according to biological data and environment data collected in the ecological protection zone, the biological data being used to generate a biological feature vector for animal and plant identification, and the environment data being used to generate an environment feature vector for environment quality evaluation;

[0241] The processing acquisition module 402 is configured to process the animal and plant information, the environment quality evaluation information, the biological feature vector, the environment feature vector and a comprehensive feature vector based on an evaluation large model, to acquire an ecosystem service function evaluation result of the ecological protection zone, the comprehensive feature vector being determined based on the biological feature vector, the environment feature vector and a biological weight and an environment weight dynamically adjusted and fused;

[0242] The generation construction module 403 is configured to generate a comprehensive feature matrix based on the animal and plant information, the environment quality evaluation information and the ecosystem service function evaluation result, and to construct a digital twin engine through deep learning based on the comprehensive feature matrix, wherein the comprehensive feature matrix is formed by fusion feature vectors of a plurality of space-time units arranged in a space-time sequence, and a fusion feature vector of a single space-time unit is formed by multi-modal feature splicing of animal and plant information, environment quality evaluation information and ecosystem service function evaluation information corresponding to the current space-time unit.

[0243] The rendering module 404 is configured to, after constructing a digital twin large model based on the digital twin engine and model static base information, render a linkage state of a species, an environment and an ecosystem service function in a visual interface in response to input of real-time data of the digital twin large model, the visual interface being a virtual reality (VR) interface or an augmented reality (AR) interface.

[0244] Optionally, the acquisition module comprises:

[0245] The generation sub-module is configured to generate the biological feature vector based on the collected biological data and generate the environment feature vector based on the collected environment data, the biological data at least including image data, audio data and positioning data of animals and plants, and the environment data at least including meteorological data and physical environment data.

[0246] The acquisition sub-module is configured to perform species identification on the biological feature vector based on a biological large model to acquire the animal and plant information in the ecological protection zone, and perform environment analysis on the environment feature vector based on an environment large model to acquire the environment quality evaluation information in the ecological protection zone.

[0247] Optionally, the processing acquisition module comprises:

[0248] The splicing determination sub-module is configured to splice the biological feature vector and the environmental feature vector based on the dynamically adjusted biological weight and environmental weight to determine the comprehensive feature vector;

[0249] The analysis generation sub-module is configured to perform comprehensive inference analysis on the biological feature vector, the environmental feature vector, the comprehensive feature vector, the animal and plant information, and the environmental quality evaluation situation based on the optimized evaluation large model to generate an ecosystem service function evaluation result.

[0250] Optionally, the generation construction module comprises:

[0251] The processing determination sub-module is configured to perform data preprocessing and time and space alignment on the multi-source data in each spatio-temporal unit to determine standard data, wherein the multi-source data in a single spatio-temporal unit comprises the comprehensive feature vector, the animal and plant information corresponding to the current spatio-temporal unit, the environmental quality evaluation situation, and the ecosystem service function evaluation result;

[0252] The extraction generation sub-module is configured to perform multi-modal feature extraction in the standard data and generate a feature vector of each modality based on the extracted features, wherein the multi-modal features at least comprise biological features, environmental features, and service function features;

[0253] The splicing acquisition sub-module is configured to acquire a weighted feature vector of each modality in the same time and space dimensions by using the dynamically adjusted weight of each modality, and splice the weighted feature vectors of each modality to generate a fusion feature vector of the current spatio-temporal unit;

[0254] The construction sub-module is configured to arrange the fusion feature vectors of a plurality of spatio-temporal units in a time sequence and a spatial grid sequence to form the comprehensive feature matrix.

[0255] Optionally, the generation construction module comprises:

[0256] The learning output sub-module is configured to perform deep learning on the comprehensive feature matrix to output a weight matrix, a response function, and a threshold rule, wherein the weight matrix comprises a plurality of coupling weights and is used to quantify the action relationship between species and environment and service function, the response function is used to describe the response relationship between ecosystem variables, and the threshold rule is used to define the critical state of the ecosystem;

[0257] The generation sub-module is configured to generate a dynamic kernel for calculating the state of the ecosystem, evaluating the trend of state change, and identifying the critical state based on the weight matrix, the response function, and the threshold rule.

[0258] The encapsulation construction sub-module is configured to encapsulate the dynamic kernel and a rule engine, and construct the digital twin engine, wherein the rule engine makes a decision response based on an output result of the dynamic kernel.

[0259] Optionally, the learning output sub-module comprises:

[0260] The learning unit is configured to input the comprehensive feature matrix into a deep learning model, learn a coupling relationship between the species variable, the environment variable and the service function variable, and output a weight matrix comprising a coupling weight between the species and the environment, a coupling weight between the species and the service function, a coupling weight between the environment and the service function, and a coupling weight of multi-element synergy.

[0261] The fitting determination unit is configured to embed a nonlinear activation layer in the deep learning model, fit a functional relationship between the species variable, the environment variable and the service function variable, and determine a response function.

[0262] The learning generation unit is configured to learn a key ecological threshold based on a loss function constraint in a training process of the deep learning model, and generate a threshold rule.

[0263] Optionally, the system further comprises:

[0264] The loading module is configured to load model static base information for constructing a three-dimensional space skeleton and basic attributes to the digital twin engine in the case of constructing the digital twin engine based on deep learning.

[0265] The processing module is configured to perform spatial registration and logical binding of the digital twin engine and the model static base information, and configure various model parameters in the digital twin engine to initialize the digital twin large model.

[0266] Optionally, the rendering module comprises:

[0267] The updating sub-module is configured to access real-time data obtained based on a data interface to the digital twin engine, drive the digital twin engine to update a model state of the digital twin large model, and the real-time data comprises real-time plant and animal information, real-time environment quality evaluation, and real-time ecosystem service function evaluation results.

[0268] The processing output sub-module is configured to process the real-time data and ecosystem historical data based on the updated digital twin large model, and output an ecosystem operation result, wherein the ecosystem operation result indicates a linkage state of the species, the environment and the ecosystem service function, and the ecosystem operation result comprises at least one of a current state of the ecosystem, a state change trend of the ecosystem, a critical state of the ecosystem, and a decision response.

[0269] The display submodule is configured to visualize the ecosystem running result output by the digital twin large model through a VR interface or an AR interface.

[0270] Optionally, the system further comprises:

[0271] The update acquisition module is configured to update a model state of the digital twin large model in response to receiving an interactive operation on the VR interface or the AR interface, and acquire an updated ecosystem running result output by the digital twin large model.

[0272] The update module is configured to update the visualized content on the VR interface or the AR interface in real time based on the updated ecosystem running result.

[0273] For the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts are referred to the part of the method embodiment.

[0274] It should be noted that the above detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0275] It should be noted that the terms used herein are only for the purpose of describing specific embodiments, and are not intended to limit the exemplary embodiments described herein. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprise" and / or "include" are used in the specification, they refer to the presence of a feature, step, operation, device, component, and / or combinations thereof.

[0276] In the above detailed description, reference was made to the accompanying drawings, which form a part of the detailed description. In the drawings, like reference numerals typically identify like components, unless the context clearly indicates otherwise. The illustrative embodiments described in the detailed description, drawings, and claims are not meant to be limiting. Other embodiments can be used, and other changes can be made, without departing from the spirit or scope of the subject matter presented herein.

[0277] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for constructing a digital twin of a terrestrial ecosystem, characterized in that, The method comprises the following steps: According to the biological data and environmental data collected in the ecological protection zone, the information of animals and plants and the environmental quality evaluation situation in the ecological protection zone are obtained, the biological data are used to generate a biological feature vector for animal and plant identification, and the environmental data are used to generate an environmental feature vector for environmental quality evaluation; Based on the evaluation large model, the animal and plant information, the environmental quality evaluation situation, the biological feature vector, the environmental feature vector and the comprehensive feature vector are processed to obtain the ecosystem service function evaluation result of the ecological protection zone, and the comprehensive feature vector is determined based on the biological feature vector, the environmental feature vector and the fusion of dynamically adjusted biological weight and environmental weight; Based on the animal and plant information, the environmental quality evaluation situation and the ecosystem service function evaluation result, a comprehensive feature matrix is generated, and a digital twin engine is constructed by deep learning on the comprehensive feature matrix; The deep learning on the comprehensive feature matrix comprises: Deep learning on the comprehensive feature matrix outputs a weight matrix, a response function and a threshold rule, the weight matrix includes a plurality of coupling weights and is used to quantify the action relationship between species and environment and service function, the response function is used to describe the response relationship between ecosystem variables, and the threshold rule is used to define the critical state of the ecosystem; Based on the weight matrix, the response function and the threshold rule, a dynamic kernel for calculating the state of the ecosystem, evaluating the trend of the state change of the ecosystem and identifying the critical state is generated; The dynamic kernel and a rule engine are packaged to construct the digital twin engine, and the rule engine makes decision response based on the output result of the dynamic kernel; wherein the comprehensive feature matrix is composed of fusion feature vectors of a plurality of space-time units arranged in a space-time sequence, and each fusion feature vector of a space-time unit is formed by multi-modal feature splicing of the animal and plant information, the environmental quality evaluation situation and the ecosystem service function evaluation result corresponding to the current space-time unit; After constructing a digital twin large model based on the digital twin engine and model static base information, in response to inputting real-time data of the digital twin large model, the linkage state of species, environment and ecosystem service function is rendered in a visual interface, and the visual interface is a virtual reality (VR) interface or an augmented reality (AR) interface.

2. The method of claim 1, wherein, The method comprises the following steps: Based on the biological data collected, the biological feature vector is generated, and based on the environmental data collected, the environmental feature vector is generated, the biological data at least include image data, audio data and positioning data of animals and plants, and the environmental data at least include meteorological data and physical environment data; Based on the biological large model, the species identification is performed on the biological feature vector to obtain the information of animals and plants in the ecological protection zone, and based on the environmental large model, the environmental analysis is performed on the environmental feature vector to obtain the environmental quality evaluation situation in the ecological protection zone.

3. The method according to claim 1 or 2, characterized in that, The evaluation large model is used for processing the animal and plant information, the environment quality evaluation condition, the biological feature vector, the environment feature vector and the comprehensive feature vector, and an ecosystem service function evaluation result of the ecological protection zone is obtained, and the ecosystem service function evaluation result comprises: The biological feature vector, the environment feature vector and the dynamic adjusted biological weight and environment weight are combined to splice vectors to determine the comprehensive feature vector; The biological feature vector, the environment feature vector, the comprehensive feature vector, the animal and plant information and the environment quality evaluation condition are comprehensively inferred and analyzed based on the optimized evaluation large model to generate the ecosystem service function evaluation result.

4. The method of claim 1, wherein, The animal and plant information, the environment quality evaluation condition and the ecosystem service function evaluation result are used to generate a comprehensive feature matrix, and the comprehensive feature matrix comprises: The multi-source data in each space-time unit is preprocessed, and the time and space of the data are aligned to determine standard data, and the multi-source data in a single space-time unit comprises the animal and plant information, the environment quality evaluation condition and the ecosystem service function evaluation result corresponding to the current space-time unit; Multi-modal feature extraction is performed on the standard data, and each modal feature vector is generated based on the extracted features, and the multi-modal features at least comprise biological features, environment features and service function features; in the same time and space dimension, the weighted feature vector of each modal is obtained by dynamically adjusting the weight of each modal; The weighted feature vectors of each modal are spliced to generate a fusion feature vector of the current space-time unit; The fusion feature vectors of a plurality of space-time units are arranged in time sequence and space grid order to form the comprehensive feature matrix.

5. The method of claim 1, wherein, The comprehensive feature matrix is subjected to deep learning to output a weight matrix, a response function and a threshold rule, and the deep learning comprises: The comprehensive feature matrix is input into the deep learning model to learn the coupling relationship between the species variable, the environment variable and the service function variable, and a weight matrix comprising the coupling weight between the species and the environment, the coupling weight between the species and the service function, the coupling weight between the environment and the service function and the coupling weight of multi-element cooperation is output; A nonlinear activation layer is embedded in the deep learning model to fit the functional relationship between the species variable, the environment variable and the service function variable to determine the response function; During the training process of the deep learning model, the key ecological threshold is learned based on the loss function constraint to generate the threshold rule.

6. The method according to claim 1 or 5, characterized in that, Further comprising: In the case of constructing the digital twin engine based on deep learning, the model static base information used for constructing the three-dimensional space skeleton and the basic attributes is loaded into the digital twin engine; The digital twin engine and the model static base information are spatially registered and logically bound, and various model parameters in the digital twin engine are configured to initialize the digital twin large model.

7. The method of claim 6, wherein, In response to inputting real-time data of the digital twin large model, the linkage state of the species, the environment and the ecosystem service function is rendered on the visualization interface, and the linkage state comprises: The acquired real-time data is accessed to the digital twin engine based on a data interface, driving the digital twin engine to update a model state of the digital twin large model, the real-time data including real-time plant and animal information, real-time environmental quality assessment, and real-time ecosystem service function assessment results; The real-time data and ecosystem historical data are processed based on the updated digital twin large model, and ecosystem operation results are output, the ecosystem operation results indicating a linkage state of species, environment, and ecosystem service function, and the ecosystem operation results including at least one of current ecosystem state, ecosystem state change trend, ecosystem critical state, and decision response; The ecosystem operation results provided by the digital twin large model are visually displayed through a VR interface or an AR interface.

8. The method of claim 7, wherein, The method further includes: In response to receiving an interactive operation on the VR interface or the AR interface, updating a model state of the digital twin large model, and acquiring updated ecosystem operation results output by the digital twin large model; Based on the updated ecosystem operation results, the visual content on the VR interface or the AR interface is updated in real time.

9. A digital twin construction system for a terrestrial ecosystem, characterized in that, It includes: An acquisition module is configured to acquire plant and animal information and environmental quality assessment in the ecological protection area according to biological data and environmental data collected in the ecological protection area, the biological data being used to generate a biological feature vector for plant and animal identification, and the environmental data being used to generate an environmental feature vector for environmental quality assessment; A processing acquisition module is configured to process the plant and animal information, the environmental quality assessment, the biological feature vector, the environmental feature vector, and a comprehensive feature vector based on an evaluation large model, to acquire an ecosystem service function assessment result of the ecological protection area, the comprehensive feature vector being determined based on the biological feature vector, the environmental feature vector, and a biological weight and an environmental weight that are dynamically adjusted and fused; A generation and construction module is configured to generate a comprehensive feature matrix based on the plant and animal information, the environmental quality assessment, and the ecosystem service function assessment result, and to construct a digital twin engine through deep learning on the comprehensive feature matrix; The digital twin engine constructed through deep learning on the comprehensive feature matrix includes: Deep learning is performed on the comprehensive feature matrix to output a weight matrix, a response function, and a threshold rule, the weight matrix including a plurality of coupling weights and being used to quantify the action relationship between species and environment and service function, the response function being used to describe the response relationship between ecosystem variables, and the threshold rule being used to define the critical state of the ecosystem; Based on the weight matrix, the response function, and the threshold rule, a dynamic kernel for calculating an ecosystem state, assessing an ecosystem state change trend, and identifying a critical state is generated. The dynamic kernel and the rule engine are packaged, the digital twin engine is constructed, and the rule engine makes a decision response based on the output result of the dynamic kernel; wherein the comprehensive feature matrix is composed of fusion feature vectors of multiple space-time units arranged in a space-time sequence, and the fusion feature vector of a single space-time unit is formed by multi-modal feature splicing of the corresponding plant and animal information, environmental quality assessment situation and ecosystem service function assessment result of the current space-time unit; The rendering module is configured to, after constructing the digital twin large model based on the digital twin engine and the model static base information, render the linkage state of the species, the environment and the ecosystem service function on a visual interface in response to input of real-time data of the digital twin large model, the visual interface being a virtual reality (VR) interface or an augmented reality (AR) interface.

Images